AffineBasedEstimator crashed when called with an existing CameraParameters.
This happens e.g. when using Stitcher in SCANS mode.
CameraraParameters is now cleared before any calculation is executed.

This test case uses a matrix with more dimensions than columns. Without
the fix in
https://github.com/opencv/opencv/pull/8448/commits/b45e784bebe318826bb48a41023db3fe1660d91e
this crashes with a segmentation fault, hangs or simply fails with wrong
values.

Stitcher will now make a working copy of the CameraParams object to avoid side effects when composing Panorama.
Makes it possible to estimate transform once and then compose multiple panoramas. Useful for setup with fixed cameras.

This issue concerns only matrices with more dimensions than columns.
See https://github.com/opencv/opencv/issues/8447

* export SVM::trainAuto to python #7224

* workaround for ABI compatibility of SVM::trainAuto

* add parameter comments to new SVM::trainAuto function

* Export ParamGrid member variables

Warping a matrix with more than 4 channels using BORDER_CONSTANT and
INTER_NEAREST, INTER_CUBIC or INTER_LANCZOS4 interpolation led to
undefined behaviour. This commit changes the behavior of these methods
to be similar to that of INTER_LINEAR. Changed the scope of some of the
variables to more local. Modified some tests to be able to detect the
error described.